Robert, See inline.
Rafal Szarecki JNCIE Routing Solution Architect Juniper Networks. +14086809604 On Feb 28, 2019, at 1:48 AM, Robert Raszuk <[email protected]<mailto:[email protected]>> wrote: Hi Rafał, It seems to me that all you are doing here is setting next hop self on ASBRs (pretty well know) + tracking this next hop based on the EBGP sessions state you are applying it to when advertising intra-domain (say to your RRs). Is there anything else new here ? Not sure if I can ack your understanding. Let me put this way - ABR insert ANH route to IGP conditionally - if session is up and EoR have been received. Note that multiple session may be tracked for given ANH, wit OR logic. Question: How are you going to deal with the case where say 3 paths received from your peering ASto a given ASBR will contain different AIGP attribute (or any other attribute for that matter) ? Will you just propagate best path with ANH and hide the information contained in other two paths to the rest of your network ? Each ASBR will propagate its best route to on-site RR and then to on-site CR (add path). So CR will run normal selection to pick best. CR shall run multipath (but necessary relaxed one). But if AIGP or other attributes differs, normal selection apply. So CR creates FIB as per best path(s). Now RR is advertising single best path to other sites. With another ANH’ (RR change BGP NH). So other sites see one path. It could happen that as result of failure this path need to be updated by other path, but since they have same BGP NH, other site routers do not need to re-program FIB. Now, update over WAN can take time. During this time previous routing information is used, but since BGP NH == ANH’ it will come to same (set) of CR on originating site. And it will have accurate up-to date FIB to send traffic out vis best ASBR(s) at this time. If at RR, best and 2nd best path have same attributes, and only diff is BGP NH, since it is set to ANH’, there is no need to send update to other sites. Kind regards, R. On Thu, Feb 28, 2019 at 7:16 AM Rafal Jan Szarecki <[email protected]<mailto:[email protected]>> wrote: GROW-ers, Here is draft describing abstract Next-Hop concept and it’s use in scale-out peering. I would appreciate reviews and comments. https://datatracker.ietf.org/doc/draft-szarecki-grow-abstract-nh-scaleout-peering/<https://urldefense.proofpoint.com/v2/url?u=https-3A__datatracker.ietf.org_doc_draft-2Dszarecki-2Dgrow-2Dabstract-2Dnh-2Dscaleout-2Dpeering_&d=DwMFaQ&c=HAkYuh63rsuhr6Scbfh0UjBXeMK-ndb3voDTXcWzoCI&r=Hjhzvcy3RXY7GgnXtof0rgOeWXlbs83hVb3_12LdlBA&m=1hGKA2FSZ8xSV4kVP14dhAHrPEiyW0unU1IECnicMVM&s=Tg02rNI9DSW_s8w8SatgtMG8FyAUmPh6Mgt1Dblywrw&e=> -- Rafal Szarecki _______________________________________________ GROW mailing list [email protected]<mailto:[email protected]> https://www.ietf.org/mailman/listinfo/grow<https://urldefense.proofpoint.com/v2/url?u=https-3A__www.ietf.org_mailman_listinfo_grow&d=DwMFaQ&c=HAkYuh63rsuhr6Scbfh0UjBXeMK-ndb3voDTXcWzoCI&r=Hjhzvcy3RXY7GgnXtof0rgOeWXlbs83hVb3_12LdlBA&m=1hGKA2FSZ8xSV4kVP14dhAHrPEiyW0unU1IECnicMVM&s=JcXO6_VFWVcFcfhlja-7rKIgMuSnb5a_sDWznzas2l8&e=>
_______________________________________________ GROW mailing list [email protected] https://www.ietf.org/mailman/listinfo/grow
